Biostatistics Weekly Seminar

Building a dose-equivalence model from a Bayesian meta-analysis of published clinical trials

Elizabeth Sigworth
Vanderbilt University Medical Center

In clinical practice, medications are often interchanged in treatment protocols when a patient has a negative reaction to their first line of care. Our motivating example looks at the taxane chemotherapy drugs paclitaxel and docetaxel, which are both associated with the adverse outcome of peripheral sensory neuropathy. Currently, patient risk of this outcome is believed to be related to cumulative taxane exposure. About 20% of patients are changed from paclitaxel to docetaxel due to infusion reactions, but there is no clear guidance on how to choose the initial dose of docetaxel given the previous paclitaxel dose, making it difficult to track cumulative exposure. No studies exist that directly explore this dose-equivalence relationship, as it would be unethical to intentionally induce neuropathy side effects in subjects, which motivates us to establish this dose-equivalence relationship using published clinical trial results. We propose a method to incorporate trials that included one or both drugs of interest via a Bayesian meta-analysis model that accounts for both within- and between-study variances. With the posterior parameter samples generated from this model, we construct a median and 95% credible interval for the equivalent dose pairs of the two drugs that are predicted to produce equal rates of an adverse outcome. Via extensive simulations, we show that this approach produces a good approximation of the true dose-equivalence relationship, across different study designs, varying levels of between-study variance, and the potential inclusion of summarized subject-level covariates in addition to study-level covariates. Comparing the performance of the meta-analysis estimate to the equivalent model fit on the underlying subject-level information that generated the meta-data, we find some efficiency loss in estimating coefficients for the summarized subject-level covariates but minimal efficiency loss in the study-level coefficients used in the dose-equivalence relationship. Finally, we present the findings of our taxane dose-equivalence model, based on data gathered from 213 published clinical trials.

Zoom (Link to Follow)
14 October 2020

Topic revision: r2 - 30 Sep 2020, AndrewSpieker

This site is powered by FoswikiCopyright © 2013-2020 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Vanderbilt Biostatistics Wiki? Send feedback